Reporting Things I Wish I Had Learned Sooner

JD Long
2018-10-26

Disclaimer

* I am sharing my personal views not the views of RenaissanceRe Holdings, Ltd. or any of its subsidiary companies.
* Thought provoking only: The materials are provided for information purposes only, are in summary form, and thus do not include all information necessary for analysis. 
* This is not consulting: No representations or warranties are give as to the accuracy or completeness of the materials and no professional advice or guidance is given by the presentation. If you want to use these ideas, professional consultation should be sought. 
* This is not financial advice: No financial advice is being given nor any recommendation to invest in any financial product. Historical data is no assurance as to future results.

Source

All data, code, and examples: CerebralMastication/Presentations

Take Pix, Dance, Ask, Tweet @cmastication

What's Happening Here?

– Obligatory XKCD

– Key Point: R Fits into Corporate Reporting

– R in Corporate Reporting Fits Better if You're Not an Asshole (empathy!)

– 3 Easy Tricks to Try Tonight to Make Your Reporting Hotter

JD Who?

Stack Overflow…

@cmastication

All Corporate Risk... All the Time

JD Who?

Second Edition, Y'all!

Speed Survey...

  • Who works at a “for profit” business?

https://xkcd.com/1205/


Assumptions:

– All time is of equal value and is interchangeable

Assumptions:

– All time is of equal value and is interchangeable

– Current frequency of reporting = future freq

Assumptions:

– All time is of equal value and is interchangeable

– Current frequency of reporting = future freq

– All activities are of equal utility (nothing sucks your will to live!)

Assumptions:

– All time is of equal value and is interchangeable

– Current frequency of reporting = future freq

– All activities are of equal utility (nothing sucks your will to live!)

– Automated and manual workflows produce the same product (errors?)

A Moment on Empathy...

A Moment on Empathy...

It looks like you're doing a data science!

Would you like me to format your integers as dates?!?

What Problem are We Solving?

  • Reproducibility
  • Multiple Users
  • Easy to alter
  • Looks right to audience
  • Useful

Structuring Reporting Workflow

– Use R Studio Project structure (or here package)

Structuring Reporting Workflow

– Use R Studio Project structure (or here package)

– Keep all files in sub directories in relative paths

Structuring Reporting Workflow

– Use R Studio Project structure (or here package)

– Keep all files in sub directories in relative paths

– Naming convention - make it sort right

Structuring Reporting Workflow

– Use R Studio Project structure (or here package)

– Keep all files in sub directories in relative paths

– Naming convention - make it sort right

– Sub Directories: make it easy to grok

Structuring Reporting Workflow

– Use R Studio Projects (or here package)

– Keep all files in relative paths

– Naming convention: sorting

– Sub Directories: easy to grok

– Passwords in environ variables

Three Homes for Code:

– Markdown File (*.Rmd) - minimal

– Sourced R script (*.R) - most here

– Package: if used more than 2 times

Tips:

– Params - Knitr YAML - For easy maintenance

– Worded package by David Gohel - Page Breaks and page orientation hacks

– Flextable package by David Gohel - Table formatting controls

– Officer package by (guess who?) David Gohel! - more table formatting

Whatsit look like?

---
title: "My Amazing Report"
output:
  worded::rdocx_document: 
     reference_docx: 10_amazing_report_template.docx
     toc: false
     toc_depth: 2
date: '2018-10-26'
params:
  main:    'mycompany 2018-10-01'
  compare: 'mycompany 2018-07-01'
---

Whatsit look like?

```{r, include=FALSE}
library(flextable)
library(our_kickass_corp_package)

my_corp_db_con <- 
   our_db_connect(Sys.getenv('MYUID'), 
                  Sys.getenv('MYPWD'))

Flextable...

Flextable Continued...

regulartable(my_cars) %>%
  color(i = 1, color = "#003e51", part = "header") %>%
  autofit %>%
  theme_zebra(odd_header = "transparent", 
              even_header = "transparent") %>%
  fontsize(size = 9, part = 'all') %>%
  border_outer(
    border = officer::fp_border(color = "gray70", 
                                width = 2),
    part = "all") %>%
  align(align = "center", part = 'header')  %>%
  align(j = 1, align = "right", part = 'body')  %>%
  align(j = 2:4, align = "center", part = 'body') %>%

Flextable Continued...

  set_formatter(
    `mpg` = function(x)
      formatC(x, 
              digits = 1, 
              format = "f", 
              big.mark = ","),
  )

Figure Tip - Set DPI & sizes:

```{r, echo=FALSE, dpi=300, fig.width=5, fig.height=2.9}

Loop Output...

Child Doc Loop:

```{r, echo=FALSE}
my_groups <- unique(mtcars$cyl)
out <- list()
for (group in my_groups){
    out = c(out, knitr::knit_child('10_sub.Rmd'))
}

Children are evaluated in the context (namespace etc.) of the parent

`r paste(out, collapse='\n')`

Hack: Excel Output

library(openxlsx)

xlfile <- 'my_junk.xlsx'
wb <- loadWorkbook(xlfile)
removeTable(wb=wb, sheet='Sheet1', table="our_table")

writeDataTable(
  wb, sheet='Sheet1', x=my_df,
  tableName="our_table", 
  startCol=3, startRow=10
)

saveWorkbook(wb=wb, file=xlfile, 
             overwrite = TRUE)

JD Long

Github: CerebralMastication/Presentations

Twitter: @cmastication Email: jdlong@gmail.com